Maybe consequence of particular steps you take when initializing the GPIOs and the timer? You could confirm this by single-stepping through that initialization, or by inserting delays between individual writes to registers.
Or maybe hardware problem, which causes also the noise on both signals?